FormFactor

The operating-leverage inflection is now demonstrated, not projected: four quarters of >30% revenue growth, 1,500bps of non-GAAP gross-margin expansion and a tripling of EPS, with the Q3 guide implying another record. FormFactor sells a per-design consumable into exactly the two nodes where AI capex is concentrated — HBM stacks and advanced-packaged foundry logic — and the balance sheet is net cash with capex now falling.

• Record Q2: revenue $258.2M (+31.9% YoY) with GAAP gross margin at 50.7% and non-GAAP at 53.3% — a 430bps sequential step and 1,480bps YoY, evidence the high-fixed-cost model converts incremental HBM/foundry volume into margin.
• Management's own framing is quantified and checkable: over the past four quarters revenue grew more than 30%, non-GAAP gross margin expanded 1,500bps, and EPS tripled.
• The Q3 guide is a guide-up, not a maintenance guide: $270M +/- $10M revenue and 54.0% +/- 1.5% non-GAAP gross margin, both above the Q2 actuals and materially above where the Street sat (~$247M / ~$0.62 EPS).
• Demand breadth improved — the release cites strength in DRAM, Foundry & Logic AND Systems, with HBM driving Probe Cards and Co-Packaged Optics driving Systems sequentially. Systems had been the soft segment; it is now a second growth vector.
• Cash generation has flipped hard: FCF of $52.6M this quarter versus -$47.1M a year ago, as capex fell from $66.3M to $9.6M with the capacity build largely behind it. H1 FY2026 FCF is $83.3M against -$40.8M in H1 FY2025.
• Fortress balance sheet: ~$334M net cash, total debt of only $11.6M, and the equity investment now contributing income ($2.2M in the quarter versus a $3.5M loss a year ago).
• Strategic positioning in Taiwan reinforced via an expanded multi-year partnership with Keystone Microtech, addressing the ecosystem where advanced-packaging test capacity is concentrated.

The equity is behaving far worse than the fundamentals — the stock is ~23% below its June 2026 level despite two consecutive record quarters, which says the market is pricing peak-cycle margins rather than doubting the quarter. The core risk is that probe cards are a short-cycle consumable tied to customer capex and design-release cadence, so a 53% gross margin printed at a demand peak is not a durable base rate.

• Price action is the loudest bear signal: $108.80 on 2026-08-03 versus $141.83 on 2026-06-24 — a ~23% decline through a period containing a large beat and a guide-up. Intra-period volatility was extreme ($83.45 close on release day, +26% the next session).
• Margin quality is flattered by one-off adjustments. GAAP gross margin of 50.7% is helped by restructuring charges falling to $4.3M from $21.5M in Q1; the $27.8M of H1 restructuring add-backs and $16.6M of stock-based compensation mean the 53.3% non-GAAP figure excludes real recurring costs.
• Factory start-up costs of $4.9M are still running through operating expenses ($11.9M in H1), so the cost structure is not yet settled and the capex trough may not hold if HBM capacity needs another leg.
• Consumable, per-design revenue means low visibility: there is no backlog disclosure and no full-year guide — only a one-quarter outlook. A pause in DRAM/HBM qualification cycles converts directly into a revenue air-pocket.
• Customer concentration is structural in probe cards (the company discloses >10% customers separately on its IR site rather than in the release), so a single memory or foundry customer's capex slip is material.
• Export-control and tariff exposure is explicitly flagged in the risk factors, and the Taiwan/China test ecosystem is exactly where the growth is being sourced.
• Even after the drawdown, the multiple embeds continued records: ~$8.48B market cap against a ~$1.03B annualized revenue run-rate and ~$260M of annualized non-GAAP net income at Q2's rate.
• The $18.0M deferred grant on the balance sheet and the CHIPS-era capacity commitments it implies carry execution conditions that are not detailed in the release.